What's new

v4.5.4 — September 2026

Fixed (Four-Digit Years When You Stop Working): The boxes for the month you stop working and the month your spouse stops working let some browsers accept a year of five or six digits, so an extra keystroke could turn 2032 into 20325 and carry it into the projection. Both boxes now take a four-digit year only.

Projections are unchanged. Nothing in this release touches the model or any figure in the report.

v4.5.3 — September 2026

Fixed (Interest On Taxable And Cash Savings Is Now Taxed): The growth on your taxable and cash savings was never taxed. It now counts as ordinary income in each retired year it is earned, which also brings it into how much of your Social Security is taxable and into the Medicare surcharge test. This mostly matters late in a plan: a household whose required minimum distributions are more than it spends can build up a large taxable balance, and the interest on it was going untaxed.

Fixed (What The Cost-Basis Figure Applies To): Your cost-basis percentage describes the taxable balance you enter. The model was applying it to every withdrawal for life, so already-taxed money put back into the account years later — leftover income, unneeded required distributions — was still taxed as if part of it were gain. The gain in the balance you entered is now used up as it is withdrawn, and money added later adds none.

Fixed (A Spouse Who Keeps Working): When one of you keeps working after the household retirement date, their pay helps fund spending. The model counted their whole salary as spendable while also adding their retirement-plan contribution to savings, so those dollars were counted twice, and it charged no Social Security or Medicare payroll tax. It now counts take-home pay — salary less payroll tax and their own plan contribution — and leaves the traditional part of that contribution out of taxed wages, as a real paycheck does. The downloaded report labels this line “Take-home pay, spouse still working”.

Disclosed (State Tax Deductions): State tax is worked out from the first dollar of income, without any state’s standard deduction or exemption credits, so it reads high — roughly $700 a year for a California couple with about $70,000 of taxable income. This is now stated on the methodology page while the per-state figures are verified.

Projections move. Most plans with taxable savings now project somewhat more tax and a lower ending balance, and some that had leaned heavily on a partly-taxed account now show less capital-gains tax. No input has changed; a saved plan simply reads under the corrected rules.

v4.5.2 — September 2026

Fixed (Return Variation In The Consultant Workspace): The model pairs your expected growth rate with how much returns vary from year to year, so the range of outcomes describes a portfolio that could exist. Until you set the variation yourself, it follows the growth rate. In the consultant workspace, changing the growth rate left the variation where it was — a household moved to a 12% growth rate kept the 12% variation that belongs to an 8% one, while the assumptions table said it followed the growth rate. The model sensitivity estimate then drew from a calmer portfolio than the one described, and read as more reassuring than it should. Variation now follows the growth rate everywhere, including in households already saved in that state; a variation you entered yourself is kept.

Fixed (Earliest Social Security Start Month): Social Security pays a retirement benefit only for a month you are 62 for the whole month, which for most people is the month after their 62nd birthday — the birthday month itself only for someone born on the 1st or 2nd. The planner accepted the birthday month for everyone. It now names the first month that applies and asks for that or later, and the what-if start-age control moves to it rather than before it.

Year-by-year projections are unchanged. The range of outcomes moves for any household whose growth rate had been changed in the consultant workspace without setting the variation.

v4.5.1 — September 2026

Fixed (Overlapping Text In The Downloaded Report): On the “What you told us” appendix of the downloaded PDF, a household with a partner and any other regular income entered — part-time work, rental income, an annuity — could see that section's heading printed directly over the bottom of the household details above it. The appendix lays its entries into two columns, and that specific combination was tall enough to trip a layout defect in the PDF engine that undercounted where the next entry belonged. The columns are now built a row at a time instead of two long independent columns, which keeps every block short enough to lay out correctly.

Projections are unchanged. Nothing in this release touches the model, the report's figures, or any figure in it — only where one heading was drawn on the page.

v4.5.0 — September 2026

New (A Password On The Backup File): A backup downloaded from the consultant workspace holds every household’s dates of birth, salaries, balances and benefit amounts, and that file travels — into a downloads folder, onto whatever cloud drive the desktop syncs to, and often into an email. You can now give it a password. The file is encrypted in your browser before it is written, so the figures are not readable in it, and opening it later asks for the same password.

Two things worth knowing. The password is never sent anywhere and cannot be recovered by anyone, including us — without it the file cannot be opened, which is what makes it worth setting. And this protects the downloaded file, not the workspace: households held in your browser are not encrypted, and neither is a backup you choose to download without a password. A backup saved before this release still opens exactly as it did, and is never asked for a password it never had.

The Download current work button that appears when a save has failed stays unprotected, deliberately. It is a recovery path, and a recovery path must not be able to fail on a forgotten password.

Projections are unchanged. Nothing in this release touches the model, the report or any figure in it.

v4.4.0 — September 2026

Fixed (Saving With The Workspace Open In Two Tabs): The consultant workspace saves every change as you make it, with no save step. Until now each save first checked that the whole stored record still matched what that tab read when it opened — so a second tab left open anywhere, on any household, was enough to fail every later save in the first one. The message offered “Retry saving”, which re-ran the same check and could never pass, leaving that tab unable to save until it was reloaded. A save now re-reads what is stored and merges only the household it changed, so work in a second household is never disturbed. A tab that is simply open follows the other tab's saves as they happen instead of drifting out of date.

Improved (When Two Tabs Change The Same Household): This is the one case that is a genuine conflict, and it is now the only one refused. Your version stays on screen and you choose: keep it, or take the version the other tab saved. Both finish. Which household a tab has open is also remembered per tab, so two tabs can sit on different households.

New (Start A Household, Delete A Household): Browsing an example stores nothing, and the header now says so. A button on the example starts a household from it, rather than a household appearing the moment you happen to change one of its figures. Saved households can also be deleted, with the name shown on the confirming button. There was previously no way to remove one.

Improved (What “Saved” Means): The household count read “Saved households”, which a tester reasonably took for a count of saves — it never moved, because renaming a household does not add one. It now reads “Households”, and the line under the household name carries the time of its last save. When a save has failed, the workspace no longer says “Saved on this device” alongside the warning that it is not.

Fixed (A Tab That Could Not Be Closed): When a save had failed, the workspace asked the browser to confirm before leaving the page — sensible protection for work that is on screen but not stored. Because a failed save could not previously be cleared, that confirmation latched: every attempt to close the tab, reload it or navigate away raised the prompt again, for as long as the tab was open. A save that fails can now be resolved, and the prompt is limited to work that exists nowhere else — downloading it ends the risk and the prompt with it, and the banner says the tab can be closed. Changing something afterwards brings the prompt back, because that change is unstored in its turn.

Improved (Choosing A Household): The household list now opens on the line beside the button that asked for it, rather than in a new row underneath that repeated the heading you had just clicked. Nothing on the page moves down to make room for it, and the list itself says when each household was last saved.

Fixed (Telling Two Households Apart): Every new household was called “Meeting household”, so a second one was indistinguishable from the first in the list until it was renamed — and identical to it if both were renamed the same. A new household now carries the name of the example it started from, with a number if you start two from the same one, and renaming it onto a name already in use says so rather than leaving you to notice. The button that creates one now says it starts a new household, because it leaves the household you were working on.

Projections are unchanged. Nothing in this release touches the model, the report or any figure in it.

v4.3.0 — September 2026

New (Today's Dollars Or Future Dollars): A switch in the section row at the top of the results page, and beside the headline, now shows every figure in the report either in today's dollars — the default, measured at today's buying power so every year compares with your budget now — or in future dollars, inflated each year at the rate you entered so a figure reads like the amount on a statement or a bill in that year. The switch applies to the whole report at once, never to one chart, so two charts can never show the same plan in different dollars. The section row now stays in view as you scroll, so the switch is within reach beside the figures further down.

Projections are unchanged: the model still calculates in today's dollars, and the future view multiplies each year's figures by that year's inflation. Amounts you entered — savings, your monthly budget, premiums — stay as you entered them and are marked as today's dollars. Shares such as the first year's draw on your savings are measured in today's dollars in both views. The downloaded report uses whichever view you were looking at and names it in the page footer. Each time you open the report it starts in today's dollars.

Fixed (Headline At A Shorter Age): With the view shortened to age 80, 85, 90 or 95, the opening sentence and the savings runway said the projection reached that age with the balance left at age 100. They now use the balance at the age they name.

v4.2.0 — September 2026

New (Taxes And Medicare Over Time): The breakdown of federal tax, capital gains tax, state tax and Medicare income-related surcharges by age was calculated for every report but only ever appeared in the downloaded PDF. It now has its own section on the results page, with a short explanation written from your own entries — when a Roth conversion window ends, the age required distributions begin for your birth year, when a long-term care cost starts, and the first year a Medicare surcharge applies — so the shape of the chart is explained rather than left to guesswork.

Improved (Today's Dollars, Stated On Each Chart): Every money chart on the results page and in the PDF now says under it that figures are in today's buying power. Until now that was stated once, in the assumptions panel, and a level bar was easy to read as a cost that never rises rather than one keeping pace with inflation.

Fixed (PDF Chart Keys): Charts in the downloaded report are captured as images, and their colour keys were never part of the image. The taxes chart had no key at all, and the funding chart's key had been drawn by hand with colours that no longer matched the bars and was missing two sources. Both keys are now drawn from the same definitions the on-screen charts use. The funding page also no longer describes an expense line the chart does not draw, or later expenses as future-dollar values.

Improved (Survivor And Pension Context): With the survivor view on, the results page says whether your spouse's Social Security claiming age still affects household income after their modeled death — it often does not, because the survivor receives the higher of the two benefits. Where one pension continues to a survivor and the other has no continuation entered, the evidence ledger now says so, including the Survivor Benefit Plan election a military pension requires.

Improved (Saving More, Close To Retirement): When fewer than five working years remain, the saving-more change now says how many, because that is why its modeled effect is small next to spending less.

Projections are unchanged: every figure added here is one the model already calculated.

v4.1.3 — September 2026

Fixed (What-If Layout On Smaller Screens): The previous release stopped the what-if controls from overlapping on a narrower screen, but did it by dropping them into two rows of two — which replaced the intended layout rather than protecting it. The four controls and the row of figures beneath them now stay as one line, and a narrower panel tightens the type instead. Restacking is reserved for panels too narrow for any readable size to hold four controls across.

v4.1.2 — September 2026

Fixed (What-If Alignment On The Report Page): The previous release aligned the what-if controls and figures, but decided when to restack them from the width of the browser window rather than the width of the panel itself. On the report page the panel sits inside an indented section and is narrower than the window suggests, so on a smaller screen the spouse's start age wrapped onto a second line and the last figure dropped below the others while the layout still believed it had room. The panel now measures itself, so it restacks at the same point wherever it appears — report page, sample plans, or consultant workspace — and the point was set from the widest content, including plans whose balances run into seven figures.

v4.1.1 — September 2026

Fixed (What-If Alignment): In the interactive what-if, the four lever boxes now start and end on the same line as each other — the try-an-age link made its label 2px taller than the others, and the spending box was 1px deeper at each end than the plain fields. Below the result, the plan outcome and the three figures beside it now sit on one shared baseline rather than the outcome riding slightly high, and the figures step down in size before they would wrap so the row stays a single line. Where the panel is too narrow to hold four levers on a line, the levers and the figures now restack at the same width instead of at two different ones.

v4.1.0 — September 2026

Fixed (The What-If Compared Two Different Ages): With a shorter cutoff selected, the interactive what-if reported your current view at that age beside the original scenario at age 100 — two different ages, presented as one comparison, so part of the difference you were reading was just the extra years. Both figures are now read at the same age, the chart marks the age they are read at, and the original scenario still shows its own age-100 figure separately. Projections are unchanged; what changed is which two numbers are set against each other.

Improved (Interactive What-If): The section now answers before it explains. Each lever states its own change against the original — spending, each Social Security start age, and the cutoff age — and the result reads as three figures at one size: what this view leaves, the change, and what the original leaves at the same age. The four paragraphs of explanation that stood between the controls and their effect now sit on the lever they explain, on hover and on keyboard focus, with the wording unchanged and still read out by screen readers. The chart is taller, keeps its side margins, and labels every gridline.

Fixed (Overlapping Figures On The What-If Chart): When the two paths finished close together, or against the top or bottom of the plot, the two dollar amounts could be drawn on the same line. They now move apart in opposite directions, and figures are drawn with a clear halo so a path crossing behind one no longer strikes through it.

The consultant workspace groups household identity and switching above the view navigation and backup actions. Rename opens beside the household name, with Done on the right. Saved households use a compact selector, and example browsing keeps Restore backup and Return to household together. Browser-storage notices remain visible; calculations are unchanged.

v4.0.4 — September 2026

The consultant workspace highlights only the selected view: Board, Review & Choices, or View report. Review & Choices remains selected while editing inputs. The active button uses a blue background and white text instead of an underline.

v4.0.3 — September 2026

Corrected the height of the savings growth-rate controls in the consultant calculator. The increase/decrease arrows retain their compact sizing, keeping the growth-rate controls aligned with their balance fields. Input values and calculations are unchanged.

v4.0.2 — September 2026

Survivor modeling now requires an explicit consultant choice. An inherited spouse-death assumption is omitted from working household projections until selected. Review distinguishes an unanswered choice from deliberate exclusion. Sample reports and saved historical comparisons retain their original assumptions; confirmed household choices are preserved.

v4.0.1 — September 2026

Healthcare badges now describe the entered premium consistently while coverage details remain available for review. Confirming no pension updates each person's badge correctly. Survivor assumptions inherited from an example are explicitly identified in review, results, and PDF source disclosures; calculation settings are unchanged.

v4.0.0 — September 2026

The consultant board now keeps separate households in this browser and continues into the existing full report and detailed calculator. Example switching preserves saved work. Each answer retains its source, including unreviewed sample values and confirmed absences.

Review & Choices identifies the remaining answers and returns after focused editing. Saved What If comparisons retain their original household inputs. Report downloads carry household source disclosures; a shorter What If cutoff is stated separately from the full age-100 projection. The board also uses age 100 and includes both pensions and current Social Security payments. Calculator rules are unchanged.

• Fixed (An Earlier Finish Shows On The Timeline Too): When one of you stops working before the other, the timeline kept showing a single shared “Work income stops” on the household’s date — on the rail of the person who had already finished. Their earlier exit was reaching the projection the whole time, ending their workplace saving sooner and lowering the balance you reach retirement with, but the picture said nothing had moved. Each person’s line now carries their own date whichever of you stops first, and it says what an earlier finish costs: no living costs are charged before your retirement date, so nothing is drawn on savings there — what ends is the saving into them. A date already in the past is left off, rather than stretching the timeline back to reach it. No figure moves; the picture now matches the figures it always had.

• Changed (One Of You Can Keep Working): The model used to retire a couple on a single date, so the moment one of you stopped, the other's pay stopped too — whatever date their own box said. For the households who plan it that way, usually to carry the health cover until Medicare, the projection removed a salary that was still arriving and covered the gap out of savings instead. Each person's work-exit date now runs their own pay. While one of you is still earning, that income appears as its own band in the funding chart, keeps their workplace saving going, and is taxed as ordinary income like any other — so it counts toward the Medicare surcharge thresholds and the pre-Medicare premium subsidy as well. The timeline shows the two dates separately instead of one shared one. This moves figures for any couple with different work-exit dates and an entered salary; a household where both stop together, or where no salary was entered, projects exactly as before.
• Fixed (One Name Per Growth Rate, Again): There are three savings balances and two growth rates: taxable and cash has its own, while tax-deferred and Tax Advantage (Roth) share the other. The shared one was announced to screen readers as "Tax-deferred growth rate", naming one of the two balances it drives — a Roth-heavy household could reasonably read that as excluding their Roth savings, and it does not. The report called the same rate "Retirement account growth rate", which does not exclude the taxable balance either, since all three are retirement savings. Both rates are now named for exactly the balances they drive:Taxable and cash growth rate andTax-deferred and Roth growth rate, on the calculator, in the report, in the assumptions panel and on the what-if comparison. No figure moves.
• Fixed (The Claiming Age Accepts What You Type): On the discussion board, "Age they claim Social Security" would not take an entry. Every keystroke was corrected into the valid range before the next one arrived, so 70 could never be reached through 7 and the box always snapped back. It now holds what you are typing until it is a real claiming age, and an answer aimed past 62 or 70 lands on the nearer of the two. No figure moves.

• Fixed (The Three Change Cards Line Up): In "What changes the picture", a card whose title ran to two lines pushed its bar lower than its neighbours', so the three bars could not be read against each other. The cards now share one set of rows, so the title, the figure, the bar and the note sit level across all three whatever wraps. The figure itself stays on one line and shrinks with the card, so a six-figure effect fits at any width. No figure moves.

• Fixed (The Timeline Is Readable Again): For most households the changes in a retirement fall in its first few years, and the timeline was spreading four decades evenly, so half a dozen labels shared a tenth of the width and overlapped. The axis now gives its width to the years that hold something and breaks before the horizon. Two changes in the same year share one dot, labels step into separate rows with a short lead line where they would otherwise touch, and changes to the whole plan sit on the year axis rather than on a person's row. The printed report draws the same layout. No figure moves.

• Changed (The Report Leads With The Answer): The results page and the printed report now open with the result in one sentence — how far the savings reach against the age you asked to plan to — with the three monthly figures behind it. It used to open with a legend and make you read down to find the answer.
• Added (Four Pictures Instead Of Four Paragraphs): Where the money comes from is drawn as income stepping up to meet your spending. How long the savings last is a runway across the years you asked about. The single changes worth discussing are cards showing how far each one moves that runway. And the market test is a hundred dots, of which the filled ones are the histories where savings were still there. Every figure is the same one the tables already carried; none of them is new arithmetic.
• Changed (Seven Stops, And A Place To Check The Working): The report's five sections became seven, so exploring a scenario and reading what would sharpen the result each get their own stop instead of being folded into a longer one. Assumptions, model rules, what is not modelled and the full calculation trace now sit in a closing section of their own, one accordion each.
• Changed (The Detail Is One Click Down, Not Gone): Each stop now shows its picture, one sentence reading it in monthly terms, and the evidence state beside it. Everything the page used to show in the open — the coverage bar, the annual figures, the formula, the source notes, the withheld cards — sits in a "Show the working" drawer under the picture, closed until you open it. The legend for the four evidence states folds into the badge on the opening result instead of taking a section of its own.
• Added (Your Timeline, Drawn): The dated changes in your plan are now drawn as a lifeline, one row for each of you, with the years between retirement and Medicare shaded and the year savings run out marked where there is one. Filled dots are your decisions; hollow dots happen because of someone's age. The written list of the same events sits underneath for anyone who wants the reasons.
• Added (Each Open Entry Names Its Document): Where the report is resting on an estimate, a default or a blank, it now says where to get the real figure — the plan statement, the account statement, the marketplace quote — rather than only naming the entry. Where the entry is a decision rather than a fact, it says there is nothing to fetch. The same line prints in the PDF's evidence ledger.
• Changed (The Printed Report Tells The Same Story): The PDF now walks the same seven stops as the page, in the same order, with the same sentence under each one. It draws the lifeline, the single changes as bars against your plan as entered, and closes with the open entries and where to get each figure. Every number comes from the same place the page reads it, so the report you print cannot disagree with the one on screen.
• Changed (The Result Comes First): The results step used to open with a step heading, three summary cards, and a header repeating the figure the result then stated again. All of that is gone. Above the result there is now one row: the way back to your inputs, the start-over control, and the download button. The four-state legend folds into the badge on the result.
• Fixed (Every Sample Household Runs In The Calculator): Five of the six sample households could be read on the demo page but stalled at step 2 if you entered them into the calculator yourself, because the salary question had never been answered for them. Each now answers it in the way that leaves its figures exactly as published, so every sample report can be reproduced and downloaded. The one exception is withheld on purpose: its claim month is blank, and that is what the report is about.
• Added (Your Report Can Use Your Name): The first step of the calculator now asks for a first name for each person, and both are optional. Enter them and the report speaks to you directly — “Dana reaches Medicare age”, “Mark's Social Security start age” — on screen and in the PDF. Leave them blank and every one of those lines reads exactly as it did before, “you” and “your spouse”. No projected figure moves either way: a name changes how the report addresses you, never what it calculates.
• Unchanged (Nothing Leaves Your Browser): A name is treated like every other input — held in your browser with the rest of the plan, used only to label your own report. Saved plans made before this release open normally with the fields empty.
• Fixed (A Sentence That Never Read Right): When the salary question was left unanswered, the calculator asked you to “Choose whether you is still earning a salary.” It now agrees with whoever it is addressing — “whether you are”, or “whether Mark is”.

• Changed (The Printed Report Is Now A Retireomics Report): The product became Retireomics on 2 September 2026 and the site moved with it, but the downloadable report had gone on calling itself the Retirement Decision Model on its cover, in its running footer, in its document properties, in its disclosures and in several sentences of explanation. It now carries the same lockup as the site: the Datum mark, RETIREOMICS, and "retirement decision model" as the descriptor beneath it. The non-affiliation and Monte Carlo disclosures name Retireomics, the running footer names the model version and retireomics.com, and the file is saved as Retireomics-Report-<date>.pdf. No projected figure moves and no engine rule changed.

• Changed (The Product Has A New Name): Kessavo is now the Retirement Decision Model — the name the printed report has carried since v2.x. No projected figure moves and no engine rule changed. The site, the report, the legal pages and the support address all move together, and the canonical address is now www.retirementdecisionmodel.com. Links to the old address redirect permanently.
• Unchanged (Saved Plans Survive): The browser key that holds a saved plan was deliberately left alone, so a plan started before the rename opens normally after it. Renaming that key would have silently discarded every saved plan on every tester's device.
• Note (New Look Still To Come): The logo, favicon and social images still carry the previous mark. The new identity is drawn but not yet exported, and lands in a later release.

• Changed (One Product Instead Of Two Tiers): The model is no longer described as a free tier with paid additions. Couples, Monte Carlo, long-term care, state tax handling and cost-basis tracking all shipped to everyone on 14 August, and several pages had gone on describing an offer that no longer existed. The site now describes a single product, with what it does and what it does not do set out at the same length on the same page.
• Changed (What The Model Will Not Do, Written As An Agenda): The limitations were previously a list of denials, which testers read as a reason not to bother. Each one now says what it means for you and what to do about it — the boundaries of the model are the agenda for a conversation with a professional, not a warning label.
• Removed (Prices No Longer Published): Pricing is not announced until launch, and the previous figure described the superseded tier structure. It has been removed from the changelog, the homepage, the product page, the FAQ, the year-by-year table inside the calculator, and from llms.txt, which had been supplying the old price to AI search engines.
• New (Privacy Policy Describes Billing Before It Exists): The policy now names Stripe as the payment processor, states that card details are entered with Stripe directly, and describes the sign-in-link model and exactly what it leaves us holding — an email address and Stripe's record of whether access is current. Disclosed ahead of the feature rather than alongside it.
• Changed (Withheld From Search And AI Indexes Until Launch): Every page now sends noindex, and AI training and answer-engine crawlers are disallowed inrobots.txt. Ordinary search crawlers are still permitted to fetch pages — a crawler has to be able to read a page to see that it should not be indexed. Reversed at launch.
• Removed (Empty Guides Page): /guides had been an unindexed placeholder since launch and has been taken down along with its footer link.
• Changed (The Sample Pages Have Their Own Design): The households index and each household's page were rebuilt to a new visual treatment — the six as cards that are deliberately not uniform, a profile and ledger for each household, their questions set as quotes, and the findings and answers as distinct cards rather than lists. The report itself is unchanged; only the pages around it were redesigned.
• Added (Six Worked Examples You Can Read End To End): Six sample households now have a full report each, at /demo. They are deliberately unalike: a couple retiring together in California, a single woman covering her own health insurance for three years before Medicare, a military pension with a survivor election, a widow living on a survivor benefit, a household that answered only half our questions, and a man whose savings run out at 81. Each page shows what the household entered, the questions they actually asked, and then the real report — computed in your browser as you read it, with the controls live. Nothing is a screenshot and nothing is withheld.
• Added (Where The Model Stops, And Who To Ask Instead): Each sample names the point at which the model runs out of road, and what to do about it — which professional, and the specific question to bring them. The widow's report cannot tell her whether switching to her own Social Security record at 70 would pay more, because it does not hold her earnings record; it says so, and points her at the SSA rather than estimating.
• Changed (The Expandable Sections Now Look Expandable): Five sections of the report hold its best material — the rule and arithmetic behind a figure, the full year-by-year table, every assumption and every place the model takes a shortcut. They were rendered as a bold line with a small plus beside it, which reads as a heading, so almost nobody opened them. Each is now a full-width row with a chevron at the end, it responds when you move the pointer over it, and a second line says what is inside — “your inputs · how the result is built · model rules · what is not modeled · the calculation trace” — so you can see whether it is worth opening before you open it.
• Fixed (A Health Subsidy Was Applied Against A Salary The Model Treated As Zero): This changes projected figures, so it is worth reading if you entered a salary and retire part-way through the projection. The premium tax credit is worked out from the previous year's income, which is correct — but earned income was not part of the income the model held, so the first retired year was priced against a working year recorded as earning nothing. Households were shown a full subsidy in that year that they would not receive. Earned income is now included in that calculation. Nothing else about tax changed. If your plan showed a large first-year subsidy, it will be smaller or gone, and the earlier figure was the wrong one.
• Fixed (A Younger Spouse Lost All Health Cover The Moment The Older One Turned 65): This changes projected figures for couples with an age gap. Health cost was worked out from the older person's age alone, so in the years between one 65th birthday and the other, the younger partner — still years from Medicare and still buying private cover — was charged nothing at all. For a two-year gap that quietly removed two years of the most expensive cover a near-retirement plan carries. Cost is now worked out per person and added together: each of you is either under 65 and buying cover, on Medicare, or neither. Plans with an age gap will show a higher cost in those years, and a lower ending balance.
• Fixed (The Same Figure Was Reported Twice With Two Different Answers): "Paths funded to age 100" appeared in two places on one report and could disagree by a percentage point, because two different tests were counting it — one counted plans that never ran out at all, the other counted plans that survived past the age. Both readings are defensible; a report may only use one. It now uses a single definition everywhere, matching the words on the page: funded through age 100 means the money lasted to the end of that year.
• Fixed (Two Ending Balances Printed On Top Of Each Other): On the what-if chart, the original and current ending balances are both drawn at the right-hand edge. When the two paths finished close together the figures overlapped into an unreadable smear — and the code meant to prevent that was causing it, nudging both labels upward so a small gap became none at all. Both figures now sit stacked below the end of the lines, in the same order as the paths they belong to, clear of the line and the marker.
• Fixed (Age Labels Under The Charts Ran Into Each Other): The funding, tax and balance-range charts printed a label for every single year of a retirement horizon — thirty-five of them across one axis, colliding into an unreadable smear at anything narrower than a wide desktop. They now print about ten evenly spaced ages, scaled to the length of the horizon rather than a fixed step, so the axis stays readable on a laptop and a tablet. No figure changes; this is what the axis says, not what the model computes.
• Changed (A Report Never Calls Someone Else's Figures Yours): The report header distinguishes three cases rather than two — a reader's own submitted inputs, an internal wireframe sample, and a published sample household. The sample pages are labelled as illustrative composite households throughout, including in the page metadata that travels when a link is shared.

• Fixed (The Timeline Reported “$0 To $0” On The Year That Changed Most): This matters if your plan runs out of savings at some point, or if your pension and Social Security already cover your spending. The timeline described the year Social Security starts by reporting what came out of your accounts — and that figure is $0 both when nothing needed to come out and when there was nothing left to take. A household short $4,406 a month before their benefit and $1,406 after was told the draw “moves from about $0 to about $0”, no change at all, while the chart beside it showed that same gap in red. The timeline now names what your income covers, what your savings actually supply, and what is left unfunded, so those two situations no longer read alike.
• Fixed (“Your Savings Carry Your Spending” Was Said To Everyone): The first stop on the timeline used one sentence for every household. If your income already covers your spending, you were told your savings were carrying it. If you had nothing saved, you were told the same thing. It now says which of the three is actually happening, and no longer announces that withdrawals begin for a household that draws nothing.
• Improved (The Evidence Ledger Lines Up): On the uncertainty page, the two tables listing what is an estimate, a default or a blank sized their columns separately, so the same headings started in different places and the pair read as unrelated tables. They now share one set of columns, matching the printed report.
• Fixed (Mistyped Addresses Now Say So): Any address that did not match a page quietly returned the homepage instead, with no indication that the page you asked for does not exist. There is now a real “page not found” page that says what happened and links to where you were probably going.

• New (Every Number Now Says How Solid It Is): Each result carries one of four states — complete, provisional, unavailable, or not applicable — with the reason named and what would firm it up. A state is never a judgement about your finances, only about the evidence behind the number. Two results are now deliberately withheld rather than approximated: comparing claiming ages needs exact dates and the month you intend to claim, and your health-coverage timeline needs a real quoted premium.
• Fixed (A Blank Was Being Read As A Zero): A monthly spending figure nobody typed, a $0 pension, an untouched 8% return and a 100% cost basis were shown at the same weight as figures you actually entered — and the printed report listed them side by side under one heading. Pension, debt and long-term care are now questions where “no” is a real answer and silence is recorded as silence. The modelling assumptions moved into their own panel you confirm, so a figure you never looked at is reported as our starting value rather than as yours.
• New (Dates, Not Just Years): Date of birth, the month you stop working, the month you intend to claim, and the date you read your balances. Social Security is worked out to the month, so a January and a December birthday in the same year are different amounts — asking for a year and comparing claiming ages from it produced an answer that looked exactly like a right one.
• New (Your Headline Is A Date You Choose): The report used to open on the balance projected at age 100, which compounds an assumed return for nearly four decades. It now opens on where you stand on a date you pick: what arrives without touching savings, what you plan to spend, and what has to come out of your accounts to close the difference — with no return or inflation forecast in it.
• New (Essential And Discretionary Spending): Spending is now two figures rather than one — what still has to be paid in a difficult year, and what you would choose to spend on top. Without the split, no page could show you which part of a gap was optional, which is the part that matters most when a year goes badly.
• New (You Pick The Comparison): A step before the results asks you to choose one thing to compare against and one uncertainty you care most about. The report runs your alternative beside your plan and describes the difference. It no longer ranks claiming ages for you or calls one of them best.
• Rebuilt (Both Reports, Five Questions): Position, Capacity, Timing, Choices, Uncertainty — one headline each, on screen and in print. The audit trail moved behind the story rather than sitting second. Every chart is kept: how spending is funded sits under Timing, the interactive what-if under Choices, and the 300-scenario stress test under Uncertainty, where it now says what it does not vary. A new page lists everything in your plan that is an estimate, a default or a blank, ordered by how much it could move the picture.
• New (The Survivor Question Is Asked): Every couple is now asked what happens if one of them dies first, and the spouse's pension is a full part of the plan. Leaving the question unanswered is recorded as unanswered rather than left out of the report entirely.
• Improved (Saved Plans Carry Forward): A plan saved before this release is migrated rather than discarded. A saved birth year becomes mid-year and is marked approximate, and anything that depends on it stays provisional until you confirm the real date.

What's next

• Save-and-compare — store multiple scenarios locally and compare them side-by-side without re-running the wizard.
• Enhanced survivor-benefit transition modeling for married households.
• Expanded state-specific marginal tax bracket rules beyond initial 13 states.